DEM and image based loess slope segmentation

An integrated method combining digital terrain analysis (DTA) and object-oriented image analysis is used to classified loess slope into six types according to surface shape. Firstly, several data layers are extracted from Digital Elevation Models (DEM). Secondly, the data layers are combined into a multi-band image. Thirdly, object-based image analysis is use for slope segmentation and classification based on the multi-band image. It is proved that the methodology is effective, and useful in geomorphological research. Then, to quantitatively depict the classification result, several landscape indices are tested. Results show that these indices can properly describe the size, shape, aggregated degree, connectivity degree of each slope patch. Our study also provides a potential way to quantitative research of loess landform from the landscape ecology point of view.